Purification and Structural Characterization of Progastrin-derived Peptides from a Human Gastrinoma ”

Several peptides derived from the gastrin-predicted preprohormone sequence were isolated from a human gastrinoma by gel permeation, anion exchange, and reverse phase chromatography. The peptides were identified and characterized structurally by a combination of radioimmunoassays, mass spectral analysis, and microsequence analysis. The largest peptide, progastrin-( 1-35) (cryptagastrin), extends from the putative processing site for the signal peptidase to the double basic residues adjacent to the amino terminus of gastrin 34. A shorter form of this peptide, progastrin-(6-35) (cryptagastrin-(6-35)), was also isolated in smaller amounts. In addition, sulfated and nonsulfated gastrin 17 amides (progastrin-(55-71)) and the glycine-extended nonsulfated gastrin 17 (progastrin(55-72)) were identified by radioimmunoassay, and their structures were confirmed by mass spectral analysis. Isolation of cryptagastrin indicates that the signal peptide of human preprogastrin contains 21 amino acid residues, and progastrin, therefore, contains 80 amino acids. There is minimal processing of the cryptic peptide preceding the sequence of gastrin 34. An amidated gastrin form larger than gastrin 34 could contain 71 amino acids. No evidence was obtained for processing that would produce gastrins containing more than 34 but less than 71 amino acid residues.

Gastrin is an important hormone that regulates gastric acid secretion and gastric mucosal growth (1,2). Hypersecretion of gastrin from gastrin-producing tumors (gastrinomas) leads to hyperplasia of the gastric mucosa and hypersecretion of acid associated with severe peptic ulcer disease, a clinical condition known as Zollinger-Ellison syndrome (3).
Gastrinomas have been the source for purification of several molecular variants of gastrin. Gregory and coworkers isolated and characterized nonsulfated and sulfated heptadecapeptide gastrins (4), "big" gastrins (5), and minigastrins (6). The latter forms of gastrin were found to contain 34 and 14 amino acid residues (7). These peptides all are carboxyl-terminal ami- dated at a phenylalanine residue and have similar biological activities as stimulants of acid secretion, although the smaller forms are cleared from the circulation more rapidly than the larger forms (8,9).
Elucidation of the cDNA structure of human progastrin has been done from mRNA extracted from gastrinoma (10) and from normal human stomach (11)(12)(13). Based on the nucleotide sequence, the structure of a 101-amino acid human preprogastrin could be deduced. This preprogastrin was predicted to contain a leader sequence of approximately 20 residues and two flanking peptides at the amino-terminal and carboxylterminal ends of gastrin 34. The carboxyl-terminal region also contained a typical amidation region consisting of glycine followed by 2 arginine residues (14). Preliminary characterization of the carboxyl-terminal progastrin peptides has been carried out by use of antibodies specific for the carboxylterminal flanking peptide (15) and for glycine-extended gastrin (16). Other region-specific antisera have been used to identify differences in posttranslational progastrin processing between gastrinoma and normal human gastric antral tissue (17).
Very little structural analysis has been done to fully characterize gastrin gene products in gastrinoma other than the amidated forms known to be biologically active. We have produced evidence that the site for signal peptidase cleavage of preprogastrin is between residues 21 and 22 of preprogastrin (18). This processing site has been confirmed by another group (19). In order to understand fully the complex processing of progastrin so that activities of various peptide products can be studied, it is essential to isolate and characterize each of the products of posttranslational processing in uiuo. The present study describes a comprehensive approach to such analyses using a combination of chromatographic separation steps combined with radioimmunoassay, mass spectrometry, and microsequence analysis.

Extraction of Gastrin and Amino-Terminal Flanking Peptides from
Gastrinoma Tissue-A 4.4-g portion of a human gastrinoma, obtained a t operation and stored a t -70 "C, was broken into small pieces, added to 20 ml of boiling water, and boiled for 15 min. After cooling on ice, ammonium bicarbonate was added to a final concentration of 0.1 M. The tissue was then homogenized with a polytron homogenizer and extracted by mild stirring a t 4 "C for 2 h. The extracted tissue was then centrifuged a t 31,000 X g for 1 h. The supernatant was then loaded onto two Sep-Pak C18-cartridges (Millipore Corp.), linked sequentially, which were washed with 20 ml of 0.1 M ammonium bicarbonate.
The concentrated tumor extract was purified by chromatography on a Vydac pH-stable Cs column in 0.1 M ammonium bicarbonate, using a gradient of 0-50% acetonitrile. The column eluant was monitored by absorbance at 214 nm. Two-milliliter fractions were col-lected and assayed for gastrin and glycine-extended gastrin immunoreactivity using region-specific radioimmunoassays (20,21). Further purification involved rechromatography of the fractions on the same column, using shallower gradients for elution.
Extraction of Glycine-extended Gastrin from Gastrinoma Tissue-In a separate isolation, 40 g of the same gastrinoma was extracted in 400 ml of boiling 0.13 M (1%) ammonium bicarbonate. The extract was centrifuged a t 3000 X g for 30 min, and 100 ml of supernatant was loaded onto five Millipore Sep-Pak cartridges combined in series. The Sep-Paks were rinsed with 10-ml portions of 10,15,20,25,30,40, and 50% acetonitrile in 0.13 M ammonium bicarbonate.
The 20 and 25% fractions containing the most glycine-extended gastrin immunoreactivity were concentrated to remove some of the organic solvent and then diluted to 50 ml with 0.13 M ammonium bicarbonate and applied to a Sephadex G-50 SF column (95 X 5 cm) equilibrated with 0.13 M ammonium bicarbonate and eluted with the same buffer. The column effluent was monitored by absorbance at 280 nm and by radioimmunoassays specific for gastrin and glycineextended gastrin (20,21). The glycine-extended gastrin and the gastrin immunoreactive peaks were separately chromatographed on fast protein liquid chromatography mono Q columns. The peptides were loaded onto columns equilibrated in buffer A (0.1% ammonium bicarbonate in 10% acetonitrile) and eluted with a linear gradient to 0.5 M NaCl in buffer A. The column effluents were monitored by absorbance a t 280 and 220 nm and by radioimmunoassays.
The immunoreactive peaks from the fast protein liquid chromatography were then loaded directly onto a Vydac pH-stable Cs column (10 mm X 25 cm) and eluted with a gradient to 20% acetonitrile in 0.1 M ammonium bicarbonate.
Radioimmunoassay-Gastrin immunoreactivity was determined by radioimmunoassay as described previously (20). Carboxyl-terminal amidated gastrin peptides were detected with antibody 1611. Glycineextended gastrin was quantitated by use of a new monoclonal antibody that reacts with glycine-extended gastrin fragments containing more than 6 residues (21). This antibody reacts very weakly with other variants of progastrin, including the completely processed amidated forms of gastrin. The glycine-extended gastrin assay utilized the synthetic peptide Tyr-Gly-Trp-Met-Asp-Phe-Gly for radioiodination and as the standard.
Amino Acid Analysis-Peptides were hydrolyzed in 6 M HC1, 10% phenol for 24 h at 110 "C. Hydrolysates were then applied to a Beckman 121 MB amino acid analyzer. Ninhydrin-derivatized amino acids were detected a t 570 and 440 nm.
Microsequence Analysis-Peptides were sequenced on a City of Hope-built gas phase microsequencer as described (22). Phenylthiohydantoin-derivatives were analyzed using the system described by Hawke et nl. (23).
Conditions for the FAB' (fast atom bombardment) analysis of gastrin and sulfated gastrin were optimized using a synthetic gastrin sample. The numerous carboxylic acid side chains with or without a sulfated tyrosine and the absence of amino groups strongly favor the production of negatively charged ions during the FAB ionization process. Therefore, the negative ion mode was essential for analysis of the picomole amounts of these peptides. Our experiments indicate that amounts in excess of 50 pmol were readily detected, provided no impurities such as involatile buffers were present to inhibit sample ionization. A polar organic solvent such as dimethyl sulfoxide or N,Ndimethylformamide was needed to effectively remove small amounts of gastrin from polypropylene tubes. Because the progastrin peptide fragments contain few acidic functional groups, they were analyzed in the positive ion mode to achieve maximal sensitivity.
Positive and negative ion FAB mass spectra were obtained using a JEOL HXlOOHF double-focusing mass spectrometer operating a t 5 kV acceleration voltage and a nominal resolution of 3000 unless otherwise noted. Sample ionization was by means of a 6-KeV Xe atom beam. Mass spectra were obtained in two different modes of operation. For broad mass range survey scans, repetitive scans of mass-assigned data were collected directly using a JEOL DA5000 ' The abbreviation used is: FAB, fast atom bombardment. data system. Scan-to-scan cycle times were typically 1 min over the mass range of 100-4000 m/z. More accurate determinations of the molecular ion species were made by accumulating in the interface of the data system several scans over a narrow (50-100 m/z) mass range about the molecular ion. Subsequent peak detection and mass assignment was done by the host computer (DEC P D P 11/73). Mass values reported are for the monoisotopic (M -H)-(negative ion spectra) or (M + H)+ (positive ion spectra) molecular ion species except for the glycine-extended gastrin sample where the data were collected at a resolution of 500, and average mass values are reported.
Samples previously collected in polypropylene microcentrifuge tubes and evaporated to dryness in a vacuum centrifuge were dissolved in 2-5 pl of dimethyl sulfoxide. Approximately 2 ~1 of the solution was added to approximately 1 pl of 10% ethanolamine in glycerol (negative ion FAB spectra) or 1 pl of dithiothreito1:dithioerythritol (5:l) (positive ion FAB spectra) on a 1.5 X 6-mm stainless steel sample stage. Excess solvent was evaporated in the vacuum lock of the direct probe inlet.

RESULTS
Gastrin and Amino-Terminal Progastrin Peptides- Fig. 1A shows the elution profile from the pH-stable Cs column for the gastrinoma extract. Two peaks of gastrin immunoreactivity were observed. Based on mass spectral analysis (Table I) of the purified samples, the earlier eluting peak was identified as sulfated gastrin 17 and the later eluting peak as nonsulfated gastrin 17. Two major absorbance peaks with no associated gastrin immunoreactivity were present in the fractions adjacent to those containing gastrin 17 and sulfated gastrin 17. The predominant peptide in each of these fractions was purified to apparent homogeneity by rechromatography on the Cs column using the same ammonium bicarbonate, acetonitrile system (Fig. 1, B and C).
Half of each of these peptides was subjected to microsequence analysis. The amino-terminal sequences obtained are given in Table 11. In order to determine their mass, 25% of     the peptide samples were analyzed by mass spectrometry. The smaller peptide gave a protonated molecular ion at m/z 3250, and the larger peptide gave a molecular ion at m/z 3904 mass (Table I, Fig. 2). These molecular weights correspond within experimental error to those of peptides extending up to the double arginine processing site just before the amino terminus of gastrin 34 (Table I, Fig. 3). Both peptides are derived from the amino-terminal region of progastrin that precedes the gastrin 34 sequence. Progastrin-(6-35) (cryptagastrin-(6-35)) could be derived from progastrin-(1-35) (cryptagastrin) by proteolytic cleavage at an arginine residue. Amino acid analysis of both peptides confirms their identity (Table 111).
Glycine-extended Gastrin-On Sephadex G-50, the major immunoreactive peak of glycine-extended gastrin eluted slightly before the major immunoreactive peak of amidated gastrin, but there was considerable overlap. Total glycineextended gastrin immunoreactivity represented approximately 20% of the gastrin immunoreactivity eluted from the G-50 column. The glycine-expended gastrin immunoreactivity was further purified by chromatography on a fast protein liquid chromatography Mono Q column. Final purification was achieved on a Vydac pH-stable C8 column, as described under "Materials and Methods." Approximately 550 pmol of the glycine-extended gastrin immunoreactivity was desalted as described under "Materials and Methods" and analyzed by FAB mass spectrometry. Upon analysis of half of the sample (at low resolution), two molecular ions were obtained. The average mass of the first was   Progastrin Processing 2155 u, in agreement with the value calculated for glycineextended gastrin 17. A second, more predominant molecular ion gave an average mass of 2171 u, corresponding to the molecular mass for glycine-extended gastrin with an oxidized methionine.

DISCUSSION
The structural characterization of gastrin offers many challenges. Its blocked amino terminus makes direct sequence analysis impossible, and its heterogeneity with respect to size, sulfation, and oxidation of methionine makes immunoreactivity studies difficult to interpret. We have used a combination of radioimmunoassay and FAB mass spectrometry to fully characterize several different forms of gastrin present in a particular human gastrinoma. Sulfated and unsulfated gastrin 17 were detected in this tissue, as well as several progastrin metabolites without known biological activities.
We report here a structural confirmation of the existence of a glycine extended precursor of gastrin 17. Previous studies (16) have demonstrated a substance with the expected immunochemical and chromatographic properties of glycineextended gastrin 17. Only immunoreactivity was used as the basis for identification (14), and no chemical replicates have been available for direct comparison.
Glycine-extended gastrin is stored in normal gastrin cells and is released into the circulation in proportions similar to those of stored forms (24,25). In cultured rat gastrin cells, amidated and glycine-extended gastrin are stored in approximately equal amounts and released in response to secretagogues in similar proportions (26). The relative amounts of glycine-extended gastrins are increased to variable extents in human disease associated with hypergastrinemia, including gastrinoma (27,28). The biological activity of glycine-extended gastrins appears to be much lower than that of amidated gastrins, although some activity has been detected on isolated gastric parietal and somatostatin cells (29). In man, a glycine-extended gastrin tetradecapeptide did not affect basal or stimulated gastric acid secretion (30).
We have also fully characterized by microsequencing and mass spectrometry the amino-terminal peptide of progastrin. This peptide corresponds to the peptide sequence between the signal peptide and the amino terminus of gastrin 34 from which 2 basic residues have been removed. Rahier et al. (31) have predicted the existence of such a peptide and have called it "Cryptic A" peptide of progastrin. We have modified this designation and have named the peptide that was completely characterized from this tumor "cryptagastrin." Fig. 3 depicts the amino acid sequence for human preprogastrin and the peptides characterized from the human gastrinoma tissue. Our results confirm that the signal sequence cleavage site is between A1aZ1-Serz2 of preprogastrin, as was also shown in the previous partial sequence in our laboratory of an amino-terminal progastrin from another gastrinoma. Desmond et al. (19) subsequently isolated and partially characterized intact progastrin from a human gastrinoma and confirmed that its amino terminus was SerZ2.
Our earlier report (18) described the co-sequence analysis of a carboxyl terminal fragment of cryptagastrin that was not separable from cryptagastrin by Cla reverse phase high pressure liquid chromatography using a 0.1% trifluoroacetic acid/ acetonitrile elution system. The two peptides were completely resolved by the use of an ammonium bicarbonate buffer system permitting microsequence analysis of the two peptides separately. Further characterization by mass spectrometry confirmed the molecular masses of these two peptides. The carboxyl terminus of both peptides is His35 of progastrin, which probably results from cleavage after Arg36-Arg37, followed by two steps of a carboxypeptidase B-like enzyme. This processing system resembles the enzymatic steps needed to process the carboxyl-terminal regions of progastrin to form gastrin 34 and gastrin 17.
The shorter amino-terminal progastrin peptide lacked the five amino acids present at the amino terminus of cryptagastrin and appeared to arise from cryptagastrin by cleavage at a single basic residue. The existence of the same two progastrin peptides from two different human gastrinomas suggests that the processing resulting in these peptides may be found in normal tissue. Development of specific radioimmunoassay for these peptides will enable investigators to confirm this presumption.
Now that the amino-terminal flanking peptides of progastrin are known, further studies can be directed at determining whether or not specific receptors exist for synthetic duplicates of cryptagastrin or of its shorter fragment. In addition, immunoassays can now be developed to localize and quantitate these peptides in normal tissues, in an attempt to gain insight into the normal biosynthetic processing of progastrin.